Learning to Predict Extremely Rare Events

Gary M. Weiss and Haym Hirsh

This paper describes Timeweaver, a genetic-based machine learning system that predicts events by identifying temporal and sequential patterns in data. This paper then focuses on the issues related to predicting rare events and discusses how Timeweaver addresses these issues. In particular, we describe how the genetic algorithm’s fitness function is tailored to handle the prediction of rare events, by factoring in the precision and recall of each prediction rule.

This page is copyrighted by AAAI. All rights reserved. Your use of this site constitutes acceptance of all of AAAI's terms and conditions and privacy policy.